rfgboost, trained in your browser

Gradient boosting with random-forest base learners (Rust + PyO3), compiled to WebAssembly and run via Pyodide — no server, no install. Categorical features are WOE-encoded by fastwoe-rs, also compiled to WASM.

trees (n_estimators): use a categorical feature (fastwoe-rs)

Booting the WebAssembly runtime…

Pinned to Pyodide 0.27 to match the wheels' ABI (CPython 3.12 / emscripten 3.1.58). Wheels are loaded locally from ./wheels/.